+$5 bonus on your first top-up

Claude API
for developers

OpenAI-compatible API proxy to Claude. Drop-in replacement - change one line of code.

quickstart.py 
from openai import OpenAI

client = OpenAI(
    base_url="https://claude.ai-platform.space/v1",
    api_key="sk-cc-YOUR-KEY"
)

response = client.chat.completions.create(
    model="sonnet",
    messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)
Get Started View Pricing

Built for developers

Any Format, Any SDK

OpenAI SDK, Anthropic SDK, LangChain, LiteLLM, curl. Any format auto-detected. 50+ model aliases.

3 Claude Models

Haiku (fast), Sonnet (balanced), Opus (powerful). All available via one API.

Real-time Streaming

Token-by-token SSE streaming. Not fake chunking - real Claude stream events.

Pay Per Token

No subscriptions. Pay only for what you use. +$5 on your first top-up.

Persistent Sessions

Multi-turn conversations with memory. Claude remembers context between requests.

Dashboard

Real-time usage stats, request logs, billing. Full visibility into your API usage.

Our authorial idea

What makes us different from Anthropic API

Three developer tools you won't find in the official SDK — built to save you hours of boilerplate.

1 · Request Builder

JSON Constructor

Visually assemble Claude requests without diving into documentation. Pick model, tune params, add system prompt, attach tools — get a copy-paste-ready JSON payload.

  • Drag & drop tools and functions
  • Validation as you type
  • Preview full curl/Python/JS snippets
2 · Guaranteed structure

JSON Normalizer

Claude often returns loosely-structured text. Our normalizer enforces your schema and guarantees a valid JSON response every single request — no more regex parsing, no retry loops.

  • Bring your JSON schema — we validate against it
  • Auto-retry if Claude deviates
  • Hours saved on parsing and error handling
3 · Coming soon

Bring your own Claude account

Plug your personal Anthropic account into our SDK and use Constructor & Normalizer on top of your own Claude subscription — for testing and non-commercial work.

  • Zero double-billing
  • All developer tools, your quota
  • Full control over your keys

Built for developers who ship products on top of Claude — chatbots, AI agents, document analysis. We solve two specific pains: unstructured LLM output and slow prototyping of requests.

Simple pricing

Pay per token. No monthly fees. No commitments.

Haiku 4.5
Fast & affordable
$1.00 / 1M input
$5.00 / 1M output
~1,000 requests per $1
  • Fastest response (~1s)
  • Simple tasks, translations
  • Git commits, summaries
Popular
Sonnet 5
Best balance
$3.00 / 1M input
$15.00 / 1M output
~200 requests per $1
  • Great quality (~2-5s)
  • Code generation, analysis
  • Default model
Opus 5
Most capable
$5.00 / 1M input
$25.00 / 1M output
~40 requests per $1
  • Deep reasoning (~5-30s)
  • Complex architecture
  • Research & debugging

All models

Prices per 1M tokens. Cache read is shown per model - it is not the same share everywhere. Models marked "on request" are opened per key.

Model Context Input Cache read Output Access
Anthropic · Claude
opusDeep reasoning, agentic coding 1M$5.00$0.50$25.00 Default
sonnetBest balance, default model 1M$3.00$0.30$15.00 Default
haikuFastest, simple tasks 200K$1.00$0.10$5.00 Default
OpenAI · GPT
gpt-6-astraMost capable GPT, 872K context 872K$10.00$1.00$50.00 On request
gpt-5.6-solHeavy reasoning 872K$5.00$0.50$30.00 On request
gpt-5.5Previous flagship 272K$5.00$0.50$30.00 On request
gpt-5.6-terraBalanced, long context 872K$2.00$0.20$12.00 On request
gpt-5.6-lunaCheapest GPT, high volume 872K$0.20$0.02$1.20 On request
Meta · Muse Spark
muse-spark-1.3Your data is NOT used for training up to 1M$1.25$0.15$4.25 On request
muse-spark-1.3-contributorCheapest chat model — because Meta trains on your prompts and completions up to 1M$0.10$0.002$0.20 On request

"On request" applies to calling a model by name. Peak-load substitution is made by us and can reach any key - see the note below.

Beyond chat

Service Model Price Access
Embeddings text-embedding-3-small $0.02 / 1M tokens Default
Embeddings text-embedding-3-large $0.13 / 1M tokens Default
Speech to text gpt-transcribe $0.0045 / minute Default
Image generation gpt-image-2 $0.006 low · $0.053 medium · $0.211 high, per image On request
Web search gpt-5.* $0.01 per search, on top of tokens On request
Realtime voice gpt-realtime-2.1-mini $10 audio in · $20 audio out, per 1M On request

Peak-load protection: if Claude is saturated, your request is served by a comparable OpenAI model instead of failing - the same price tier, honest model name in the response. If OpenAI is exhausted too, the third link is Meta muse-spark-1.3-contributor: it costs far less and is billed by the model that actually answered, but it TRAINS ON YOUR PROMPTS AND COMPLETIONS - such a response is marked with the X-Fallback-Training header. This substitution is made by us, so it reaches every key regardless of access settings. The Meta link can be switched off, but today that is a service-wide setting, not a per-key one.

Get Started - +$5 on First Top-up